THE MEASURED MOBILE-WATER CONTENT OF AN UNSATURATED SOIL AS A FUNCTION OF HYDRAULIC REGIME

To account for observations of preferential solute transport through soil, increasingly models are used in which the total water content of the soil (theta, m3 m-3) is partitioned into an essentially mobile phase (theta(m)) and an apparently immobile fraction (theta(im)). However, few methods exist for measuring this separation in the field. Here we use the recently proposed, disc permeameter technique. Following infiltration with a tracer, measurements of the resident solute concentration directly under the disc are used to infer the theta(m)-theta(im) partitioning. For Manawatu fine sandy loam in situ, we applied this technique at the three pressure heads (h(o)) of -20, -40 and -150 mm, in order to deduce the influence of the hydraulic regime on the soil water fraction theta(m)/theta that appears to be actively involved in solute transport during infiltration. When a depth I of between 15 and 25 mm of KBr tracer was added to soil already wet by pure water to h(o), the measured mobile fraction theta(m)/theta rose from 0.41 at -20 mm, through 0.50 at -40, to 0.64 at the most unsaturated head of -150 mm. Thus, less evidence of preferential solute transport was recorded with decreasing h(o). The spatial distance between the preferential paths was observed to range from 20 to 150 mm, the separation increasing and the pathways becoming more diffuse with decreasing h(o). Depthwise dispersion of the invading solute thus increased with h(o). At the two higher heads, when I almost-equal-to 80 mm of KBr was allowed to infiltrate, the theta(m)/theta, inferred from the resident concentration observed directly under the disc, now also became 0.65. For h(o) = -40 mm, the measured rise in the resident concentration under the disc, with I, could be predicted using a dispersivity, lambda, of 20 mm in the approximation provided by the 1-D form of the convective dispersion equation. When 15-25 mm tracer was applied directly to initially dry soil (theta almost-equal-to 0.3), capillary forces drew the invading solute from the disc with much less dispersion, such that the resident concentration under the disc rose more rapidly with I. Now theta(m)/theta was found to be virtually 0.65 at all the heads. In several experiments at h(o) = -40 mm, ethanol was used as the solvent for the tracer. No change in the measured theta(m)/theta was observed. Thus hydrophobicity was deemed not to be a factor in our measurement of theta(m)/theta being consistently about 0.65. An attempt was made to parameterize the diffusive-exchange scheme that the Coats and Smith (1964) model, taken from petroleum engineering, proposes as the link between the mobile and immobile domains. However, our observations at h(o) = -20 mm suggest that no single mass-transfer coefficient xi can describe this solute exchange. Over the first few days, a xi value of 0.5 day-1 seemed reasonable, but over the next fortnight there appeared no further interdomain exchange of solute between the two domains.